test: increase timeouts on arm #1366

silverwind · 2015-04-07T18:48:39Z

Related: #1343

The static factors are still subject to change, and I think I'll look into benchmarking to obtain them dynamically. ARMv8 gets no special treatment as I think it won't need it. The ARMv6 factor is probably too low for some boards. My RPi does fine with around 2-3x, but it is overclocked and has a pretty fast SD.

If someone could start off a CI run with this PR, I'd be grateful as this should green out CI for ARMv6 and ARMv7.

silverwind · 2015-04-07T18:49:27Z

test/parallel/test-debug-signal-cluster.js

@@ -51,7 +51,7 @@ function onNoMoreLines() {

 setTimeout(function testTimedOut() {
  assert(false, 'test timed out.');
-}, 6000).unref();
+}, common.platformTimeout(3000)).unref();


This timeout seems to have been chosen too generous, so I reduced it.

silverwind · 2015-04-08T14:10:00Z

@bnoordhuis could you start off a CI run with this PR? I need it to determine if the factors need more tweaking.

bnoordhuis · 2015-04-08T14:15:39Z

@silverwind https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/465/

silverwind · 2015-04-08T16:21:38Z

Results are looking quite promising. I've rebased my branch to include silverwind@92ab3c0 which should fix two --- TIMEOUT --- cases and increased the timeout on one last problematic test. This will only leave one failing test on the iojs-armv7-wheezy bot, which is certainly not timeout related:

=== release test-child-process-exit-code ===
Path: parallel/test-child-process-exit-code
assert.js:88
  throw new assert.AssertionError({
        ^
AssertionError: 'SIGILL' === null

@bnoordhuis would appreciate another CI run!

(also I think I messed up this PR with my last rebase)

bnoordhuis · 2015-04-08T16:52:32Z

@silverwind You should probably sort out the rebase before I start another CI run, it looks like a wrong merge base now.

silverwind · 2015-04-08T17:25:19Z

@bnoordhuis branch fixed.

bnoordhuis · 2015-04-08T17:28:30Z

@silverwind https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/468/

silverwind · 2015-04-08T19:02:16Z

Quite satisfied so far. Results:

iojs-raspbian-wheezy-pi1 passing for the first time
iojs-armv7-ubuntu1401 passing (did already pass before)
iojs-raspbian-wheezy-pi2 would've passed if it didn't crash in an unrelated test:

=== release test-deprecation-flags ===
Path: sequential/test-deprecation-flags
Command: out/Release/iojs /home/iojs/build/workspace/iojs+any-pr+multi/nodes/iojs-armv7-wheezy/test/sequential/test-deprecation-flags.js
--- CRASHED ---

iojs-raspbian-wheezy-pi2 (armv6) also would've passed but ran into a --- TIMEOUT --- (seems 3066f2c isn't enough for that test)

 === release test-stringbytes-external ===
 Path: parallel/test-stringbytes-external
 Command: out/Release/iojs /home/iojs/build/workspace/iojs+any-pr+multi/nodes/iojs-raspbian-wheezy-pi2/test/parallel/test-stringbytes-external.js
 --- TIMEOUT ---

@bnoordhuis please review.

bnoordhuis · 2015-04-08T21:28:10Z

test/common.js

+  armv7: 2.0
+};
+
+exports.platformTimeout = function (ms) {


Minor style issue: function(ms)

bnoordhuis · 2015-04-08T21:57:02Z

Left some comments.

silverwind · 2015-04-08T22:17:05Z

In addition to above comments, I'll work on fixing that last timeout from iojs-raspbian-wheezy-pi2 by adding a armv6 specific factor to tools/test.py.

silverwind · 2015-04-09T01:03:22Z

@bnoordhuis good for another CI run.

iojs-raspbian-wheezy-pi2 should go blue now, and with a bit of luck, all four bots should succeed.

rvagg · 2015-04-09T05:43:50Z

FYI, the persistent failure of parallel/test-debug-port-cluster on the ARMv8 machine was because there was a long running process there from a previous test run that was blocking the debug port that test was trying to use. I've killed that and now:

So woohoo! I think this means we have solid ARMv8 support now doesn't it? I thought OpenSSL 1.0.2 was a blocker for us on ARMv8 but obviously that was addressed, by @bnoordhuis I guess. Unless anyone has a reason not to, I'm going to push forward trying to get some more server resources to figure out a release build strategy for ARMv8/AARCH64 binaries.

silverwind · 2015-04-09T09:16:27Z

@rvagg 🎉 nice, that means armv8 works pretty well already. While you're at the CI, I think the ARM bots could use some relabeling. You mentioned iojs-raspbian-wheezy-pi2 is armv7, but it's actually armv6. I suggest including the architecture name for all bots, to avoid confusion.

shigeki · 2015-04-09T09:38:13Z

😄 Glad to see a blue light on armv8! Wait for a moment for ssl.

bnoordhuis · 2015-04-09T09:50:13Z

@silverwind https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/472/

shigeki · 2015-04-09T09:55:26Z

Woops, I'll fix this.

shigeki · 2015-04-09T09:56:16Z

Sorry I missed the issue. I've just abort my job. So 472 will go on right now.

rvagg · 2015-04-09T10:41:58Z

@silverwind derp! I have an origial pi1 switched with the pi2 and they are reporting themselves as such, in the process of rectifying now

silverwind · 2015-04-09T11:19:24Z

Almost fully blue 😃

The iojs-raspbian-wheezy-pi2 failure looks to be another unrelated issue:

=== release test-net-settimeout ===
Path: parallel/test-net-settimeout
events.js:141
      throw er; // Unhandled 'error' event
            ^
Error: read ECONNRESET
    at exports._errnoException (util.js:749:11)
    at TCP.onread (net.js:538:26)
Command: out/Release/iojs /home/iojs/build/workspace/iojs+any-pr+multi/nodes/iojs-raspbian-wheezy-pi2/test/parallel/test-net-settimeout.js

@bnoordhuis please review so I can land this.

bnoordhuis · 2015-04-09T11:42:10Z

LGTM

rvagg · 2015-04-09T11:44:17Z

https://jenkins-iojs.nodesource.com/job/iojs+any-pr+multi/473/ should be with the right machines running in the right place

silverwind · 2015-04-09T13:03:13Z

Looks iojs-raspbian-wheezy-pi1 is still in the make stage after 1hr20, which is different than before. I won't let that CI run hold up this PR now, but will check the results later (I expect that armv6 job to take 5-6hrs, because that's what it takes on my RPi1).

This commit introduces platform-specific test timeouts for the ARM architectures. ARMv6 is notoriously slow so gets very large timeouts on both the timeout value for each test, as well as certain problematic individual tests. ARMv7 and ARMv8 also get slightly increased headroom. PR-URL: #1366 Fixes: #1343 Reviewed-By: Ben Noordhuis <[email protected]>

silverwind · 2015-04-09T13:11:03Z

Merged in 7049d7b

Fishrock123 · 2015-04-09T13:46:34Z

@silverwind iirc, ccache is running on those machines, so it should be less than if you are running without.

silverwind · 2015-04-09T18:40:20Z

Little Pi still going strong after 7hrs:

https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/473/nodes=iojs-raspbian-wheezy-pi1/

(I think it's safe to assume there's no cache involved here)

rvagg · 2015-04-11T07:06:56Z

it's because I changed the env of this and pi2 when you notified me they were around the wrong way - so the cache has to start again with new env vars influencing the build .. over time it should get much better

Notable changes: * C++ API: Fedor Indutny contributed a feature to V8 which has been backported to the V8 bundled in io.js. SealHandleScope allows a C++ add-on author to seal a HandleScope to prevent further, unintended allocations within it. Currently only enabled for debug builds of io.js. This feature helped detect the leak in #1075 and is now activated on the root HandleScope in io.js. (Fedor Indutny) #1395. * ARM: This release includes significant work to improve the state of ARM support for builds and tests. The io.js CI cluster's ARMv6, ARMv7 and ARMv8 build servers are now all (mostly) reporting passing builds and tests. - ARMv8 64-bit (AARCH64) is now properly supported, including a backported fix in libuv that was mistakenly detecting the existence of `epoll_wait()`. (Ben Noordhuis) #1365. - ARMv6: #1376 reported a problem with Math.exp() on ARMv6 (incl Raspberry Pi). The culprit is erroneous codegen for ARMv6 when using the "fast math" feature of V8. --nofast_math has been turned on for all ARMv6 variants by default to avoid this, fast math can be turned back on with --fast_math. (Ben Noordhuis) #1398. - Tests: timeouts have been tuned specifically for slower platforms, detected as ARMv6 and ARMv7. (Roman Reiss) #1366. * npm: Upgrade npm to 2.7.6. See the release notes (https://github.com/npm/npm/releases/tag/v2.7.6) for details.

silverwind reviewed Apr 7, 2015
View reviewed changes

silverwind added the test Issues and PRs related to the tests. label Apr 7, 2015

silverwind force-pushed the arm-test-timeout branch from 3deca2c to cb025c1 Compare April 8, 2015 17:21

silverwind added 2 commits April 8, 2015 19:23

test - increase timeouts on ARM

8855832

Increase timeout in test-child-process-fork-net2

0c2210b

silverwind force-pushed the arm-test-timeout branch from bd0f345 to 0c2210b Compare April 8, 2015 17:24

silverwind changed the title ~~test - increase timeouts on ARM~~ test: increase timeouts on arm Apr 8, 2015

bnoordhuis reviewed Apr 8, 2015
View reviewed changes

test/common.js

armv7: 2.0

};

exports.platformTimeout = function (ms) {

Copy link

Member

bnoordhuis Apr 8, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Minor style issue: function(ms)

silverwind added 3 commits April 9, 2015 01:13

Simplify arch check, style

30a1882

Add a super-slow test mode for armv6

77f952c

Fix syntax

e4a5192

silverwind force-pushed the arm-test-timeout branch from 8750c84 to e4a5192 Compare April 9, 2015 00:00

process.arch

d577d19

silverwind closed this Apr 9, 2015

rvagg mentioned this pull request Apr 11, 2015

Release proposal: v1.7.0 #1396

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

test: increase timeouts on arm #1366

test: increase timeouts on arm #1366

silverwind commented Apr 7, 2015

silverwind Apr 7, 2015

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

silverwind commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

shigeki commented Apr 9, 2015

bnoordhuis commented Apr 9, 2015

shigeki commented Apr 9, 2015

shigeki commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

bnoordhuis commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

silverwind commented Apr 9, 2015

Fishrock123 commented Apr 9, 2015

silverwind commented Apr 9, 2015

rvagg commented Apr 11, 2015

test: increase timeouts on arm #1366

test: increase timeouts on arm #1366

Conversation

silverwind commented Apr 7, 2015

silverwind Apr 7, 2015

Choose a reason for hiding this comment

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

bnoordhuis Apr 8, 2015

Choose a reason for hiding this comment

bnoordhuis commented Apr 8, 2015

silverwind commented Apr 8, 2015

silverwind commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

shigeki commented Apr 9, 2015

bnoordhuis commented Apr 9, 2015

shigeki commented Apr 9, 2015

shigeki commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

bnoordhuis commented Apr 9, 2015

rvagg commented Apr 9, 2015

silverwind commented Apr 9, 2015

silverwind commented Apr 9, 2015

Fishrock123 commented Apr 9, 2015

silverwind commented Apr 9, 2015

rvagg commented Apr 11, 2015